Bayesian Fusion for Threat Detection Throughput
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing detection systems for targeted materials at security checkpoints face challenges in achieving high throughput while minimizing false positives, as they often require scanning entire objects and lack intelligent fusion of data from multiple detection apparatuses.
Innovation Solution
An integrated detection system that scans entire objects with a first threat detection apparatus to identify suspect regions, which are then targeted by a second apparatus for more focused scanning, combining preliminary and present risk value data using a Bayesian likelihood function to produce final threat probability data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a second screening system scans only the suspect region identified by the first system, then throughput is improved, but measurement precision deteriorates due to lack of intelligent fusion of independent information
Solution Approach 1:
The patent merges independent information from multiple detection apparatuses by combining their respective risk values through statistical processing. The first apparatus scans the entire object and identifies suspect regions, while the second apparatus scans only those regions. The risk values from both apparatuses are then combined using a Bayesian likelihood function to produce a final integrated risk value, thereby merging the benefits of selective scanning with accurate threat assessment.
Solution Approach 2:
The system implements feedback by using the output from the first detection apparatus (suspect region identification and preliminary risk values) to guide the scanning strategy of the second apparatus. The second apparatus focuses its scanning on regions identified as suspicious by the first apparatus, and the results are fed back through statistical processing to generate the final threat assessment. This feedback loop enables optimized throughput while maintaining accuracy.
2Measurement precision
If conventional detection systems scan entire objects, then measurement precision is maintained, but productivity deteriorates due to low throughput
Solution Approach 1:
The patent segments the scanning process into two distinct phases performed by different detection apparatuses. The first apparatus performs a comprehensive scan of the entire object to identify suspect regions, while the second apparatus segments its scanning effort to focus only on those identified suspect regions. This segmentation allows the system to maintain detection accuracy while significantly improving throughput by avoiding redundant scanning of non-suspect areas.
3Reliability
If multiple detection apparatuses are used in sequence, then reliability is improved through reduced false positives, but device complexity increases
Solution Approach 1:
The patent applies universality by using the same statistical processing and Bayesian likelihood function methodology across both detection apparatuses. This unified approach allows the system to handle data from multiple different scanner types (e.g., CT scanner and QR scanner) in a consistent manner, reducing the complexity of integrating diverse systems while maintaining reliability through reduced false positives.
Data Source
AI summary
An integrated detection system that includes a first threat detection apparatus and a second threat detection apparatus is provided. The first threat detection apparatus may identify one or more areas within an item of baggage that may contain threats. Example threats include, but are not limited to, explosives, weapons, illegal drugs, and hazardous matter, among others. The remaining areas are in theory deemed clear of threats. The second threat detection apparatus may be configured to inspect only the suspect areas of the item of baggage that was previously identified by the first threat detection apparatus. This improves throughput and lowers the false positive rate. A method for intelligently fusing the independent information obtained by the first and second threat detection apparatuses is also provided.